10. Variants

A major feature of Mosaic is the ability to view, query, and analyze genetic variants from individual cases as well as across cohorts. This section describes the variants table, how to read the annotations for a variant, how to launch external resources, how to discuss variants with your team, the variant watchlist, and the filtering tools used to narrow a large variant list down to the handful that matter for a case.

The same variants tools are available in both projects (a single case) and collections (a cohort). In a project, you are working with the variants of one case; in a collection, the same table and filters operate across every individual in the cohort. See Collections for notes specific to cohort-level variant analysis.

10.1. The Variants Table

The Variants page is reached by selecting Variants from the left menu. By default it contains a set of computationally prioritized variants for the project. These variants have been prioritized based on their likelihood of causing rare disease and have been filtered on population allele frequency, known pathogenicity, predicted impact, and phenotype (where phenotype information is available). When HPO terms are present for the case, Exomiser is also used to prioritize the variants. The table is initially loaded in chromosome order.

A banner above the table reports how many variants are currently loaded (for example, 431.5k variants loaded). Because the default view is deliberately prioritized rather than exhaustive, an information control labelled Why aren't there more? explains which prioritization and filters are currently in effect. Adjusting the filters (described in 10.7) changes which variants are loaded. A See SQL link is also available for users who wish to inspect the exact query underlying the current view.

Each row represents a variant. The default columns include:

• Watchlist star — click to add or remove the variant from the project's Variant Watchlist.
• Chr — the chromosome, with a left-arrow control used to expand the full annotations view for that variant.
• Gene Name — the gene(s) the variant falls in.
• Genotype — compact icons showing the genotype of each sample in the family. Family members are labelled (for example, Pr for proband, Fa for father, Mo for mother), and the accompanying symbol indicates zygosity (for example, homozygous, heterozygous, or reference). This lets you read the inheritance pattern at a glance.
• Exomiser Rank — the priority rank assigned by Exomiser when phenotype information is available.
• HGVSc / HGVSp — the variant described at the coding (cDNA) and protein levels.
• Gene Consequence and Gene Impact — the predicted molecular consequence (for example, missense, intron, non_coding) and its impact category (for example, MODERATE).
• ClinVar — the ClinVar clinical significance, shown as a colored label (for example, Pathogenic, Likely_benign, Uncertain_significance).
• HPO Labels — phenotype labels associated with the gene.
• Allele Frequency — the population allele frequency (for example, from gnomAD).

The columns displayed can be changed using the Select Annotations button. To sort by any column, click the column header; clicking again reverses the sort order. Check boxes at the left of each row let you select one or more variants and act on them through the Actions button, and the Launch App button opens a connected application (such as a genome browser) for the selected variants.

10.2. Selecting Annotations

The Select Annotations button (next to Filters) controls which annotation columns appear in the table. Mosaic carries a large set of annotations for every variant — population frequencies, computational predictions, clinical significance, gene-level information, and more — and Select Annotations lets you show the ones relevant to your current task and hide the rest. This keeps the table readable while preserving access to the full annotation set when you need it.

10.3. The Full Annotations View

To see everything Mosaic knows about a single variant, click the left-arrow control in the Chr column for that row. This expands the full annotations view for the variant.

The full annotations view brings together, in one place:

• Variant identity — the genomic position and allele change (for example, 3:58159593 A > G), and the genotype of each family member (Proband, Father, Mother).
• General Info — the gene, the HGVSc and HGVSp descriptions, the ClinVar significance, quality-control status, a pedigree diagram, and the gnomAD allele frequency shown for both genome and exome data.
• HPO Terms — phenotype terms associated with the gene, including a count of all gene-associated HPO terms.
• OMIM — a link through to OMIM disease information for the gene, where available.
• gnomAD — population genetics detail, including allele frequency, allele count, allele number, number of homozygotes, population-maximum allele frequency, and the maximum PolyPhen and SIFT scores, reported separately for genome and exome datasets.
• Computational Predictions — in-silico pathogenicity and impact scores, including AlphaMissense, EVE, MutScore, REVEL, SpliceAI Max, CADD (raw and Phred), and CCR.
• Collection Information — when the project belongs to a collection, a summary of which other individuals in the cohort carry the variant, their genotype, and their affected status.

From this view you can also add the variant to the Variant Watchlist and open the External Browsers menu (see below).

10.4. External Browsers and Websites

Mosaic links out to the external resources clinicians and analysts commonly consult when interpreting a variant. Click the square-with-arrow (External Browsers) icon on a variant's row, or use the External Browsers menu in the full annotations view, to open that specific variant in an external resource.

Available resources include ClinVar, MARRVEL, GTEx, the gnomAD Browser, the UCSC Genome Browser, VarSome, dbSNP, OMIM, PubMed, UniProt, and CPAM. Selecting a resource opens it in a new tab, pre-populated with the variant or gene, so you can move from Mosaic to a reference database without re-entering coordinates.

10.5. Variant Conversations

Any variant can be discussed with your team directly in Mosaic, keeping the interpretation history attached to the variant itself rather than scattered across email. To start a variant conversation, open the full annotations view for the variant and scroll to the Variant Conversations section at the bottom. Click Create Conversation to begin a thread.

Variant conversations behave like all other conversations in Mosaic: collaborators can be notified with @mentions, the thread can be watched for notifications, and it can be pinned to the Project Home. See Communication for the full set of conversation tools.

10.6. The Variant Watchlist

The Variant Watchlist is the set of variants that have been flagged as worth tracking for a case — for example, candidate variants under active review. Click the star icon on any variant row (in the table or the full annotations view) to add it to, or remove it from, the watchlist.

The watchlist appears on the Project Home so that a case's candidate variants are immediately visible on entry, and it can be recalled at any time as a filter on the variants table (see Watchlist under pre-defined filters). This makes the watchlist a convenient working shortlist that travels with the case.

10.7. Variant Filtering

The strength of the variants table is the ability to filter a very large variant list down to a clinically relevant subset. Filtering is reached through the Filters button above the table.

Opening Filters reveals two routes:

• A set of pre-defined filters that can be applied with a single click (described in 10.8).
• Links to the Advanced Filters and HPO Term Filters panes, where filters can be built and combined in detail (described in 10.9 and 10.10).

Filters can be combined: applying several filters together returns only the variants that satisfy all of them. Use Search Filters to find a specific filter by name, Clear Filters to reset, and Search Variants to apply the current filter set and reload the table.

10.8. Pre-defined Filters

Pre-defined filters provide one-click access to the queries used most often in analysis. They are grouped within the Filters pane and include:

• Exomiser Prioritized — All Candidates and Top Candidates, based on Exomiser's phenotype-driven ranking.
• Variant Sets — Watchlist, to show only the variants on the project's Variant Watchlist.
• Known Pathogenicity — ClinVar P/LP (pathogenic or likely pathogenic) and ClinVar P/LP/VUS (also including variants of uncertain significance).
• Phenotype Association — HPO Overlaps, which loads variants in genes associated with the case's HPO terms (see 10.10).

Additional pre-defined filters, such as inheritance-pattern filters (for example, De novo and Autosomal Recessive), are also available. Selecting a pre-defined filter applies it immediately; you can then refine it further using Advanced Filters.

10.9. Advanced Filters

Selecting Advanced Filters opens the Variant Filters pane, which contains a comprehensive set of filters that can be applied individually or in combination. Filters are organized into groups, and a Search Filters box helps locate a specific one. The main groups include:

• Default Filters — Region, Genes, Sample Genotypes, Collection Genotypes, Sample Attributes, Variant Length, Variant Type, and Variant Set.
• gnomAD (Exomes and Genomes) — population-frequency filters such as Allele Frequency, PopMax AF, GroupMax AF, Allele Count, Allele Number, Homozygotes, and Hemizygotes, together with predictor scores such as SIFT Max, PolyPhen Max, PhyloP, and Pangolin.
• Exomiser — Exomiser Gene Variant Score, Exomiser MOI (mode of inheritance), Exomiser Rank, and Exomiser Contributing Variant.
• Computational Predictions — in-silico scores such as AlphaMissense, REVEL and others.
• ClinVar — ClinVar Significance (Pathogenic, Likely pathogenic, Uncertain, etc)
• Variant Review — case-management fields such as Inheritance Pattern, Variant Diagnostic Status, Lab Classification, Condition, Associated Diagnosis, and Assigned Variant Consequence.

Most numeric filters share a common layout: a Min and Max field to define the range, an Include empty values? checkbox (see the note below), and an Annotation version dropdown. When you have configured the filters you want, click Search Variants to apply them, or Close to leave the pane.

Tip: Many annotations — for example ClinVar significance and gnomAD frequencies — are updated over time, and Mosaic retains multiple versions. The Annotation version dropdown lets you filter against a specific release or against the default/latest version. This is what makes reproducible, point-in-time analysis possible, and is the same mechanism that underpins ClinVar Reanalysis.

Tip: Many variants have no recorded value for a given annotation — for example, a rare variant may be absent from gnomAD entirely. Checking Include empty values? keeps these variants in the results rather than excluding them. This is important when filtering on population frequency: a disease-causing variant is often absent from population databases, so excluding empty values could remove exactly the variants you are looking for.

10.9.1. Gene Name Filters

To restrict the table to variants in a particular gene, open Advanced Filters, click Genes > Genes, and type the gene name (for example, SCN8A). You can add several genes to the same filter.

You can also load a saved Gene Set (panel) by choosing Genes > Gene Sets and selecting the set. This populates the filter with every gene in the set, so a clinical panel can be applied in a single step. Gene Sets are created and managed under Settings & More > Genes; see Genes for details.

10.9.2. Genotype Filters

The Sample Genotypes and Collection Genotypes filters let you require specific genotypes in specific individuals — for example, requiring that the proband is heterozygous while both parents are reference, to look for de novo candidates, or requiring homozygosity in an affected child for a recessive hypothesis. Collection Genotypes extends the same idea across a cohort. Combining genotype filters with the inheritance-pattern pre-defined filters is a common way to test specific genetic hypotheses for a case.

10.9.3. Population Allele Frequency Filters

To keep only variants below a chosen population frequency, open the relevant Allele Frequency filter (for example, gnomAD Allele Frequency) and enter a Min and Max — for instance, a Min of 0 and a Max of 0.05 to retain variants seen at no more than 5% in the population. To keep variants that are absent from gnomAD, check Include empty values? (see the note above). Related frequency filters — Hemizygotes, Homozygotes, Allele Count, PopMax AF, and others — work the same way and can be combined.

10.9.4. Computational Impact Filters

To filter on in-silico impact prediction scores, open the relevant predictor (for example, AlphaMissense) and enter Min and Max values to keep variants whose score falls in your chosen range. Predictors available include AlphaMissense, REVEL, SIFT, PolyPhen, SpliceAI, CADD, and others. As with all annotation-based filters, an Annotation version can be specified.

10.9.5. ClinVar Significance Filters

To filter on ClinVar clinical significance, open the ClinVar Significance filter and select the values you want to keep — Pathogenic, Likely_pathogenic, Uncertain_significance, Likely_benign, and Benign. By default the filter uses the latest ClinVar version available in Mosaic, but an earlier version can be chosen using the Annotation version dropdown.

10.10. HPO Term Filters

Phenotype-driven filtering links the case's clinical features to the genes most likely to be relevant. Open Filters > Advanced Filters and select the HPO Term Filters tab.

There are several ways to build the list of phenotype terms to filter on:

• Search and add individual terms. Use the Search box to find a term (by name or HPO identifier) and click the + button to add it to the HPO Terms To Filter list.
• Add all of the proband's terms. Click the Pr button in the top bar to load every HPO term attached to the proband at once. There are also controls to add terms for other samples and to paste a list of terms.

Once terms are loaded, the Min. overlaps box sets the minimum number of phenotype terms a gene must be associated with before its variants are kept. Increasing this value makes the filter more specific — for example, requiring a gene to be linked to at least two of the patient's phenotypes. To remove a loaded term, click the – button next to it.

For a quick, one-step phenotype filter, the HPO Overlaps pre-defined filter (under Phenotype Association) loads all variants in genes associated with the project's HPO terms without manually assembling the list.

10.11. Suggested Workflow

The tools above can be combined into a straightforward case-analysis workflow. A typical approach is:

1. Confirm the proband's phenotype is recorded as HPO terms so that phenotype-driven prioritization is available.
2. Start from a pre-defined filter — for example, Top Candidates, ClinVar P/LP, or HPO Overlaps — to get an initial shortlist.
3. Refine with Advanced Filters: tighten the population allele frequency, require a plausible inheritance pattern using genotype filters, and narrow to a relevant gene panel using a Gene Set.
4. Open the full annotations view for promising variants, consult external resources as needed, and add candidates to the Variant Watchlist.
5. Record discussion and decisions in variant conversations so the interpretation history stays attached to the case.

Warning! The variants returned by Mosaic's prioritization and filters are intended for research purposes only and to support expert review, not to replace it. Filter thresholds, annotation versions, and prioritization settings all affect which variants are shown. Clinical interpretation and reporting decisions remain the responsibility of the qualified professionals reviewing the case.